vk_cooperative_matrix_perf is a sample/benchmark demonstrating performance of using the VK_NV_cooperative_matrix Vulkan extension, and the associated GL_NV_cooperative_matrix GLSL extension.
The benchmark queries the supported matrix multiply sizes and precisions from the Vulkan implementation, and runs a couple different shaders at various tiles sizes and reports the performance in teraflops. All matrices are row-major in memory.
Running this application requires an NVIDIA Turing GPU, and a recent driver that supports the VK_NV_cooperative_matrix extension (available from https://developer.nvidia.com/vulkan-driver).
Modifying and rebuilding the shaders requires a glslangValidator.exe with GL_NV_cooperative_matrix support. This can be built from source using https://github.com/KhronosGroup/glslang/pull/1701, and will be included in an upcoming SDK release.